metadata
dataset_info:
features:
- name: input_values
sequence:
sequence: float32
- name: attention_mask
sequence:
sequence: int32
- name: pixel_values
sequence:
sequence:
sequence:
sequence: float32
- name: labels
sequence: float64
splits:
- name: train
num_bytes: 8861928600
num_examples: 1325
download_size: 2134854714
dataset_size: 8861928600
configs:
- config_name: default
data_files:
- split: train
path: data/train-*